Liposome-Mediated Chemotherapeutic Delivery is Synergistically Enhanced by Ternary Lipid Compositions and Cationic Lipids

Abstract

Most small molecule chemotherapeutics must cross one or more cellular membrane barriers to reach their biochemical targets. Owing to the relatively low solubility of chemotherapeutics in the lipid membrane environment, high doses are often required to achieve a therapeutic effect. The resulting systemic toxicity has motivated efforts to improve the efficiency of chemotherapeutic delivery to the cellular interior. Toward this end, liposomes containing lipids with cationic headgroups have been shown to permeabilize cellular membranes, resulting in more efficient release of encapsulated drugs into the cytoplasm. However, the high concentrations of cationic lipids required to achieve efficient delivery remains a key limitation, frequently resulting in toxicity. Towards overcoming this limitation, here we investigate the ability of ternary lipid mixtures to enhance liposomal delivery. Specifically, we investigate delivery of the chemotherapeutic, doxorubicin, using ternary liposomes that are homogenous at physiological temperature but have the potential to undergo membrane phase separation upon contact with the cell surface. This approach, which relies upon the ability of membrane phase boundaries to promote drug release, provides a novel method for reducing the overall concentration of cationic lipids required for efficient delivery. Our results show that this approach improves the performance of doxorubicin by up to 5-fold in comparison to delivery of the same drug by conventional liposomes. These data demonstrate that ternary lipid compositions and cationic lipids can be combined synergistically to substantially improve the efficiency of chemotherapeutic delivery in vitro.

Graphical Abstract

INTRODUCTION

Overcoming the cellular plasma membrane barrier is essential for efficient delivery of therapeutics to intracellular targets. The primary strategy used to address this challenge has been to design membrane permeable therapeutics, which must be soluble in both the hydrophobic environment of the plasma membrane and the aqueous environment of the cytoplasm.¹ Meeting these disparate constraints complicates therapeutic design, frequently limiting efficacy. For example, chemotherapeutic drugs such as gemcitabine,² cytarabine,³ and cisplatin,⁴ all suffer from low membrane permeability. Even doxorubicin, a chemotherapeutic that is soluble in both membrane and aqueous environments,⁵ has a limited transport rate across the membrane.⁶ As a result of these and other barriers to delivery, clinically effective doses of doxorubicin and many other chemotherapeutics are associated with substantial systemic toxicity.^{7, 8}

One strategy for reducing systemic toxicity is to encapsulate chemotherapeutics within nanoparticles, which are known to localize in solid tumors.⁹ However, encapsulated drugs must ultimately be released from the particle in order to reach their therapeutic targets. The requirement for liposomal release can limit the performance of drugs. For example, encapsulating doxorubicin in liposomes reduces its efficacy by 10-fold when tested in cell culture.¹⁰ This loss in efficacy largely negates the benefits associated with improved localization of liposomes in solid tumors.¹¹ As a result, liposomal formulations of doxorubicin and other chemotherapeutics have not significantly improved the clinical success of chemotherapy.¹²

Many therapeutic liposomes consist primarily of a binary mixture of cholesterol and phospholipids with saturated acyl chains.^{9, 13, 14} This formulation minimizes the permeability of drugs across the membrane, helping to promote drug loading and stable encapsulation. However, this same stability creates a substantial barrier to therapeutic release, as described above. Ternary lipid mixtures consisting of saturated-tail phospholipids, unsaturated-tail phospholipids, and cholesterol have the ability to separate into co-existing liquid-ordered and liquid-disordered phases,^15–17 a phenomenon which has been extensively studied by biophysicists.^18–20 Lipid phase boundaries are permeable to small molecules, creating a potential mechanism for efficient release.^21–27 However, the permeability of phase boundaries can also lead to unstable drug encapsulation and premature release.^{28, 29} These challenges have largely prevented the use of ternary lipid mixtures in delivery applications. How can we take advantage of the ability of ternary lipid mixtures to facilitate drug release while maintaining stable formulations?

Ternary lipid mixtures are characterized by their mixing temperature, Tm.¹⁵ When the ambient temperature is below Tm, the membrane separates spontaneously into liquid ordered and liquid disordered phases. Above Tm, all components are miscible, producing a single homogenous liquid phase. Tm is a function of the relative concentrations of the three components. Increasing the relative concentration of unsaturated lipids raises the Tm, while increasing the relative concentration of saturated lipids has the opposite effect.³⁰ Interestingly, when a membrane comprised of a ternary composition approaches another surface, suppression of local membrane fluctuations can drive a substantial increase in Tm, leading to local membrane phase separation at temperatures for which the membrane would normally be homogenous. This phenomenon, which has recently been demonstrated using several different systems,^31–33 suggests an opportunity to exploit the advantages of ternary lipid mixtures for therapeutic delivery. Specifically, ternary liposomes with Tm values below physiological temperature should be homogenous during delivery, leading to stable drug encapsulation in circulation. However, upon contact with the cell surface, the potential of the liposomal membrane to undergo local phase separation could promote drug release, improving the efficiency of delivery.

Based on this rationale, we investigated the potential of ternary lipid mixtures to promote delivery of doxorubicin. We found that a ternary lipid mixture consisting of lipids with uncharged head groups does not greatly improve delivery in comparison to conventional binary formulations, likely owing to insufficient interaction between these liposomes and the negatively charged cell surface. Evidence from other groups has shown that cellular lipid rafts help to facilitate interactions between cells and liposomal drug carriers.^{34, 35} We speculate that the alternative scenario, will also enhance interactions. Therefore, to promote interaction between liposomes and cells, we investigated delivery of doxorubicin encapsulated within ternary liposomes of the positively charged lipid, DOTAP (1,2 dioleoyl--3-trimethylammonium-propane).^36–38 We observed that these liposomes provided a substantial delivery advantage over liposomes with a binary lipid composition that was supplemented with the same amount of DOTAP. Our results suggest that ternary lipid mixtures can be combined with cationic lipids to create novel materials with the potential to significantly improve the efficiency of chemotherapeutic delivery.

RESULTS AND DISCUSSION

Loading doxorubicin into liposomes of binary and ternary compositions.

We chose our lipid composition based on the well-studied ternary lipid system consisting of DOPC (1,2 dioleoyl-sn-glycero-3-phosphocholine), which has unsaturated acyl chains, DPPC (1,2 dipalmitoyl-sn-glycero-3-phosphocholine), which has saturated acyl chains, and cholesterol.¹⁵ The capacity of this composition to separate into co-existing liquid-ordered and liquid-disordered phases has been well documented.¹⁵ The specific lipid composition we investigated consisted of 24 mol% DOPC, 33 mol% DPPC, 5 mol% PEG2000-DPPE (1,2 dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-[Methoxy(polyethylene-glycol)-2000]), and 38 mol% cholesterol (Figure 1A). We will refer to this composition as the “ternary composition” as it consists primarily of DOPC, DPPC, and cholesterol. Here, pegylated lipids were included to reduce nonspecific interactions between liposomes and cells. This lipid composition is expected to have a Tm value of approximately 31°C.¹⁵ To evaluate the phase behavior as a function of temperature, we used established protocols³⁹ to form Giant Unilamellar Vesicles (GUVs) of this ternary composition. To enable visualization of lipid phase behavior, we included a small amount (0.3 mol%) of a fluorescently labeled lipid, Texas Red-DHPE (Texas Red 1,2-dihexadecanoyl-sn-glycero-3-phosphoethanolamine), which is known to partition preferentially to the liquid-disordered phase.⁴⁰ At room temperature, 25°C, Texas Red-DHPE in the ternary composition vesicles was visibly concentrated in a hemispherical domain when the vesicles were observed using confocal fluorescence microscopy, Figure 1B. As expected, both phases appear to be liquids, as evidenced by the relatively smooth contact line between them. Here, a smooth line of contact indicates the ability of tension along the phase boundary to control the shape of the boundary, which is a property of liquids, analogous to the rounding of oil droplets suspended in water.⁴¹ The area fraction of the phase separated domain on the GUV surface was approximately 43% ± 3% (standard deviation) of the total surface area of the GUV, consistent with our previous findings¹⁹ as well as with images in the literature.^{14, 15} At 25°C, 95% ± 13% (standard deviation) of all GUVs were phase separated. Upon heating these vesicles several degrees above the expected Tm to physiological temperature, 37°C, 0% were phase separated as evidenced by a uniform distribution of Texas Red-DHPE over their surfaces, Figure 1C.

Figure 1. Characterization of liposomes with binary and ternary compositions.

(A) A ternary phase diagram displaying the regions of liquid immiscibility observed at room temperature (25°C).¹⁵ L and S represent liquid and solid phases, respectively, and L_o and L_d represent liquid-ordered and disordered phases, respectively. The ternary composition is composed of 24 mol% DOPC, 5 mol% PEG2000-DPPE, 33 mol% DPPC, and 38 mol% cholesterol. The binary composition is composed of 5 mol% PEG 2000-DPPE, 35 mol% DPPC, and 60 mol% cholesterol. The data points are superimposed over the published phase diagram for the DOPC/DPPC/cholesterol system.¹⁵ (B) Confocal fluorescence images of binary (top) and ternary (bottom) GUVs labeled with 0.3 mol% Texas Red-DHPE. The left column shows 2-dimensional confocal slices and the right column shows 3-dimensional projections of the GUVs. All scale bars correspond to 5 μm. (C) Confocal fluorescence images of a ternary composition GUV at 25°C and 37°C. The scale bars correspond to 5 μm. A bar chart on the right displays the average fraction of vesicles that are phase separated at both 25°C and 37°C for the ternary composition. Error bars indicate standard deviation among 10 different confocal fluorescence images, each of which contained multiple vesicles (Supplement Figure S1). (D) A chart depicting the percent of initial doxorubicin remaining after overnight dialysis of encapsulated and unencapsulated doxorubicin. A red arrow highlights the drastic reduction in percent of initial doxorubicin remaining after dialysis for unencapsulated drug. The black and white squares represent individual doxorubicin loading trials, and the gray bars represent the average of the two trials. (E) Intensity based size distributions of unloaded and doxorubicin loaded liposomes, from dynamic light scattering (DLS). Binary composition liposomes averaged 126 ± 3 nm in diameter before loading and 145 ± 4 nm after loading, and ternary composition liposomes averaged 104 ± 1 nm in diameter before loading and 109 ± 4 nm after loading.

For comparison to the ternary composition, we also prepared GUVs consisting of 35 mol% DPPC, 60 mol% cholesterol, and 5 mol% PEG2000-DPPE, Figure 1A. We will refer to this composition as the “binary composition” as it consists primarily of DPPC and cholesterol, similar to the composition of conventional therapeutic liposomes.⁹ We observed that vesicles having this binary composition were uniformly labeled by Texas Red-DHPE. This uniform labeling was expected, since these vesicles consisted of a single, homogenous phase in which the lipid dye is soluble.

We next sought to encapsulate doxorubicin within liposomes of both the ternary and binary compositions. To concentrate doxorubicin within liposomes we used a transmembrane ammonium sulfate gradient, previously shown to drive encapsulation of amphipathic weak bases, such as doxorubicin.⁴² To generate the transmembrane ammonium sulfate gradient, liposomes were first extruded through a filter with a 100 nm average pore diameter. Extrusion was performed in an ammonium sulfate solution, pH 5.4, leading to entrapment of the solution within the liposomes. The liposomes were then dialyzed in phosphate buffered saline (PBS), pH 7.4, to remove unincorporated ammonium sulfate buffer and exchange the surrounding medium. At this point, the liposome interior contained ammonium sulfate, (NH₄)₂SO₄, which dissociates into sulfate anions, SO₄^2-, ammonia, NH₃, and protons, H⁺. Ammonia is permeable to the membrane and is therefore lost to the surrounding bath, while the sulfate anions and protons are retained inside the liposome. Retained protons drive a sharp reduction in the pH of the vesicle lumen. Once the transmembrane pH gradient was established, liposomes were incubated with a doxorubicin-containing solution for one hour at 40°C to achieve encapsulation of the drug within the vesicle lumen. Due to its membrane permeability, doxorubicin is able to passively diffuse across the liposome membrane and into the lumen. As doxorubicin enters, it becomes protonated in the vesicle lumen, where it precipitates with the sulfate anions to form doxorubicin-sulfate aggregates.⁴² These aggregates are impermeable to the lipid membrane, driving accumulation of the drug within the liposome. These doxorubicin-loaded liposomes were then dialyzed overnight in PBS, pH 7.4, to remove unencapsulated drug. Finally, we calculated the percentage of initial doxorubicin that remained within the liposomes after overnight dialysis, Figure 1D. The percentage of initial doxorubicin remaining was well above the amount remaining after dialysis of unencapsulated, free drug, demonstrating the ability to encapsulate doxorubicin within liposomes of binary and ternary compositions. For further characterization, the liposomes were examined via dynamic light scattering (DLS) before and after doxorubicin loading. DLS measurements indicated that the diameter of the liposomes remained similar in size during drug loading and that liposomes did not aggregate, Figure 1E. These results indicate that drug loading had little impact on the physical size of the liposomes.

Impact of ternary lipid composition on delivery of doxorubicin.

We next examined the impact of liposome-encapsulated doxorubicin on cell viability. We compared liposomes of the ternary and binary compositions to determine the extent to which the ternary composition promotes drug release and delivery, as discussed above. To address this question, we plated monolayers of about 30,000 cells 24 hours prior to incubation with one of the following formulations: (i) free doxorubicin, (ii) doxorubicin encapsulated within binary liposomes, and (iii) doxorubicin encapsulated within ternary liposomes. The compositions of these liposomes were the same as those characterized in Figure 1. For each liposome formulation and for free drug, cells were treated with increasing equivalent doxorubicin concentrations of 10, 50, and 100 μM as measured by peak doxorubicin absorbance at 480 nm (see Methods). Cells treated with free drug were incubated with solubilized doxorubicin diluted in cell culture media for a total of 24 hours at 37°C. Cells treated with the liposomal drug formulations were incubated with encapsulated doxorubicin diluted in cell culture media for two hours at 37°C, Figure 2A. After two hours, the media containing liposomal doxorubicin was removed, the cells were washed, and fresh cell culture media was added. Cells were placed back in the incubator at 37°C for an additional 22 hours prior to analysis.

Figure 2. Influence of a ternary composition on liposomal delivery of doxorubicin.

(A) A pictorial representation of doxorubicin-loaded liposome delivery to HeLa cells. (B) Brightfield images of untreated HeLa cells (left), cells treated with 100 μM free doxorubicin (middle) and 100 μM doxorubicin-loaded, ternary liposomes (right). Nonviable cells are round and clumped while viable cells are well spread and lighter in color. All scalebars represent 50 μm. (C) Representative flow cytometry histograms of untreated cells (blue); cells treated with 100 μM of doxorubicin-loaded, ternary liposomes (green); and cells treated with 100 μM of free doxorubicin (red). The increase in violet dead cell stain fluorescence as shown by the second histogram peak indicates the population of nonviable cells for each condition. (D) Percentage of nonviable HeLa cells after treatment with increasing concentrations of free doxorubicin (red); doxorubicin-loaded, ternary liposomes (green); and doxorubicin-loaded, binary liposomes (blue). All data points were measured by the violet dead cell stain viability assay, in which three independent trials, as denoted by dots, of at least 5,000 cells per trial are plotted.

Following incubation, cells were imaged under widefield microscopy to visualize cell morphology and viability. The majority of untreated cells were well-spread over the substrate, indicating viability, whereas the majority of cells treated with 100 μM free doxorubicin were rounded, indicating that they were nonviable, Figure 2B. In contrast, images of the population of cells treated with 100 μM doxorubicin encapsulated within ternary vesicles included substantial numbers of both viable and nonviable cells, Figure 2B. An amine reactive violet stain that is permeable only to dead cells was used to quantify the fraction of nonviable cells resulting from exposure to each doxorubicin formulation. Staining of nonviable cell populations led to a rightward shift in flow cytometry histograms of violet stain fluorescence, Figure 2C. Results from this flow cytometry assay show that delivery of 100 μM of unencapsulated doxorubicin (red) led to a dramatic increase in the percentage of nonviable cells when compared to populations of cells that received no drug treatment (blue). These data correspond to the representative widefield images, where 100 μM free doxorubicin showed nonviable cells as the majority and untreated showed viable cells as the majority. However, upon delivery of 100 μM doxorubicin encapsulated in ternary liposomes, only a small increase in the percentage of nonviable cells was observed (green). This intermediate effect of liposomal doxorubicin delivery on cell viability was consistent with the results of widefield imaging in Figure 2B.

Based on the results of the flow cytometry experiments, Figure 2D plots the percentage of nonviable cells following treatment with increasing doses of doxorubicin, where the drug was administered either unencapsulated (red), encapsulated in binary liposomes (blue) or encapsulated in ternary liposomes (green). As anticipated, the fraction of nonviable cells increased with exposure to increasing concentrations of unencapsulated doxorubicin, consistent with previous studies.⁴³ Notably, the overall fraction of nonviable cells was substantially reduced in populations of cells exposed to liposomal doxorubicin. Specifically, the fraction of nonviable cells in populations exposed to solutions containing 100 μM of unencapsulated doxorubicin was approximately 70%, while populations exposed to doxorubicin encapsulated within binary and ternary liposomes contained 15% and 25% of nonviable cells, respectively, Figure 2D. These results are consistent with previous studies, which have suggested that liposomal encapsulation substantially reduces the effectiveness of chemotherapeutic agents.¹⁰ The slight increase in cytotoxicity of doxorubicin-loaded ternary liposomes relative to doxorubicin-loaded binary liposomes suggests that use of a ternary composition mildly enhances drug release, Figure 2D. While there is significant improvement in cell death between ternary and binary formulations at 100 μM doxorubicin concentration (p < 0.05), the superior performance of unencapsulated doxorubicin indicates that the liposomal membrane still acts as a substantial barrier to efficient delivery. Therefore, we sought to explore whether we could further improve cytotoxicity of liposomal doxorubicin to approach that of free drug.

Encapsulation of doxorubicin in binary and ternary liposomes supplemented with DOTAP.

The ability of ternary membranes to undergo contact-mediated phase separation relies upon close contact between the liposome and another membrane surface.^31–33 Therefore, one possible explanation for the inability of ternary compositions to enhance doxorubicin delivery is the lack of adequate contact between liposomes and the cell surface. Based on this rationale, we next supplemented both the ternary and binary liposome compositions with a low concentration of the cationic lipid, DOTAP, a component which is known to promote interaction with the negatively charged cellular surface.^{36–38, 44} Notably, this approach builds upon our previous results, which showed that membrane phase separation can locally concentrate DOTAP, promoting intracellular delivery of macromolecules.¹⁹

To examine the impact of DOTAP on delivery of doxorubicin, we formulated ternary and binary GUVs containing increasing molar percentages of DOTAP. As in Figure 1, Texas Red-DHPE enables visualization of the liquid-disordered region of the GUV membrane. The images in Figure 3A show cross sections of GUVs containing 0 mol%, 3 mol%, 5 mol%, and 10 mol% DOTAP, respectively. The ternary and binary GUVs containing 0 mol% DOTAP are repeated from Figure 1B for comparison. As expected, Texas Red-DHPE partitions favorably to the liquid-disordered phase of the GUV membrane for the ternary compositions, indicating that the membrane separated into two immiscible liquid phases at room temperature. In contrast, the lipid dye was distributed evenly over the surfaces of GUVs with the binary composition, confirming that the membrane consisted of a single homogenous phase.

Figure 3. Incorporation of DOTAP in binary and ternary liposomes.

(A) Confocal fluorescence microscopy slices of ternary (top) and binary (bottom) GUVs containing 0 mol%, 3 mol%, 5 mol%, and 10 mol% DOTAP. Ternary compositions were composed of 5 mol% PEG2000-DPPE, 33 mol% DPPC, 38 mol% cholesterol. Each formulation contained 24 mol% unsaturated lipids using the following ratios of DOTAP to DOPC: (i) 3 mol% DOTAP/21 mol% DOPC, (ii) 5 mol% DOTAP/19 mol% DOPC, (iii) 10 mol% DOTAP/14 mol% DOPC. Binary compositions were composed of 5 mol% PEG2000-DPPE, 60 mol% cholesterol and the following ratios of DOTAP to DPPC: (i) 3 mol% DOTAP/32 mol% DPPC, (ii) 5 mol% DOTAP/30 mol% DPPC, (iii) 10 mol% DOTAP/25 mol% DPPC. The disordered lipid domains were labeled with Texas Red-DHPE. All scale bars represent 5 μm. (B) Bar chart depicting the area fraction of the disordered lipid domain for each of the DOTAP mol% formulations. Error bars represent standard deviation of at least 10 GUVs. (C) A bar chart displays the average fraction of vesicles that are phase separated at both 25°C and 37°C for the ternary composition containing 3 mol% DOTAP. Error bars indicate standard deviation among 10 different confocal fluorescence images, each of which contained multiple vesicles. Below, are confocal fluorescence images of a GUV of ternary composition containing 3 mol% DOTAP at 25°C and 37°C. The scale bars correspond to 5 μm. (D) Bar chart depicting the percentage of nonviable cells as measured by the flow cytometry violet dead cell stain viability assay upon delivery of unloaded vesicles for each binary and ternary DOTAP liposome formulation. 375 μg of total lipid, or the maximum amount of lipid used to deliver the highest dose of doxorubicin in later studies, was delivered for each formulation. Error bars represent the standard deviation of three independent trials of at least 5,000 cells per trial. (E) Pictorial representations of ternary GUVs at 25°C labeled with Texas Red-DHPE, and their corresponding interactions with negatively charged lectin protein labeled with ATTO-488. Confocal image cross sections reveal no co-localization of lectin and the disordered domain for liposomes containing no DOTAP, but a strong co-localization for liposomes containing DOTAP in the formulation. All scale bars correspond to 5 μm.

We then measured and compared the area fraction of the Texas Red-DHPE labeled, liquid-disordered domain for ternary GUVs containing 0 mol%, 3 mol%, 5 mol%, and 10 mol% DOTAP, Figure 3B and Supplement Figure S2. In each case, the liquid-disordered domain occupied just over 40% of the membrane surface at 25°C, regardless of DOTAP content. The consistent size of the liquid-disordered domain as DOTAP concentration increases suggests that these relatively low concentrations of DOTAP can be incorporated into the ternary liposome composition without substantially altering its overall phase behavior. We acknowledge that membrane phase behavior in micron-sized GUVs may be somewhat different than phase behavior in vesicles of 100 nm diameter, possibly resulting in non-negligible changes to the membrane phase diagram. However, even in these smaller vesicles, the liquid disordered domain should still consist of about 20,000 lipids. This is a large molecular ensemble that can be reasonably described as a continuum. Therefore, it seems unlikely to us that the membrane phase behavior would be dramatically different in these smaller vesicles. Further, at 25°C, 79% ± 27% (standard deviation) of ternary vesicles containing 3 mol% DOTAP were phase separated. Upon heating to physiological temperature, 37°C, 0% of vesicles were phase separated, Figure 3C. The phase transition of these liposomes at 37°C indicates that most DOTAP-containing liposomes will be homogenous prior to contact with the cellular surface, as was the case for the formulations in Figure 1, which lacked DOTAP.

We next evaluated the cytotoxicity of the various liposome compositions, with and without DOTAP, in the absence of doxorubicin. Here we used the assay developed in Figure 2. Specifically, we extruded ternary and binary membranes containing 0 mol%, 3 mol%, 5 mol%, and 10 mol% DOTAP through a filter with 100 nm diameter pores. We then delivered these unloaded vesicles to HeLa cells, using the same total amount of lipid as was used to deliver the highest doses of doxorubicin in Figure 2. The cells were incubated with the liposomes for two hours, after which the cells were washed with fresh media and placed back in the incubator at 37°C for an additional 22 hours, as in Figure 2A. Following incubation, we evaluated the cells using the flow cytometry-based viability assay described above (see methods). Upon delivery of unloaded vesicles, we found that the liposome formulations themselves had minimal impact on cell viability, and there was no clear correlation between cell viability and DOTAP content, Figure 3D. These results indicate that relatively low concentrations of DOTAP are not substantially toxic to cells in the chosen formulations.

Owing to its unsaturated acyl chains (18:1), we would expect DOTAP to partition preferentially to the liquid-disordered membrane phase, along with Texas Red-DHPE. To test this assumption, we examined the partitioning of a fluorescently labeled anionic protein, lectin, on the surfaces of our GUVs. Lectin has an isoelectric point of about 5.1,⁴⁵ such that a net negative charge is expected at neutral pH. Owing to the net positive charge of DOTAP, we would expect lectin to be attracted preferentially to DOTAP-enriched regions of the membrane surface. We incubated our GUVs labeled with Texas Red-DHPE in a solution of 2 μM lectin protein labeled with ATTO-488. Upon imaging the GUVs using confocal fluorescence microscopy, we observed that ternary GUVs without DOTAP did not attract the lectin protein as indicated by the lack of colocalization of the ATTO-488 fluorescent label with the GUV membrane, Figure 3E. In contrast, we observed that ternary GUVs containing DOTAP did attract the negatively charged lectin as indicated by clear colocalization of ATTO-488 with the liquid-disordered portion of the membrane, Figure 3E. To verify our observations, we also evaluated lectin protein binding to binary GUVs both with and without DOTAP and found similar results. The DOTAP-containing GUVs appeared to attract the lectin in a homogenous distribution, while the binary GUVs without DOTAP did not attract lectin at all (Supplement Figure S6). These results demonstrate that inclusion of DOTAP substantially increases electrostatic interactions with the membrane surface. Further, these results suggest that, if the membrane is induced to phase separate, DOTAP will become concentrated in the liquid-disordered membrane phase.

Ternary liposomes containing DOTAP recognize anionic surfaces.

We next sought to probe the interactions between anionic surfaces and DOTAP-containing liposomes of both ternary and binary liposome compositions. In these experiments, GUVs with ternary and binary compositions, each of which contained 3 mol% DOTAP, were dispensed onto glass coverslips and imaged using confocal fluorescence microscopy (see methods). Borosilicate glass surfaces carry a mildly negative charge in aqueous solutions of neutral pH owing to the dissociation of silanol groups.⁴⁶ Images were taken of GUVs at the coverslip surface and several micrometers above the coverslip in the z-direction as shown in Figure 4. The interface between ternary GUVs and the coverslip displayed a bright background punctuated by dark, rounded spots, suggesting that separation of the membrane into liquid ordered and liquid disordered phases occurred locally at this interface, Figure 4A (left). The line profile above the images displays the normalized fluorescence intensity along the dotted white line across the GUV. At the coverslip surface, this profile shows a plateau in intensity at the site of contact with the GUV. In contrast, at a section well above the coverslip surface, the line profile shows peaks in intensity that represent the edges of the vesicle in the confocal image slice. Comparing the intensity of the plateau to the intensity of the peaks, the relative intensity greatly decreases in the image taken above the coverslip surface, Figure 4A (right). This substantial reduction in fluorescence intensity suggests that the majority of the liquid disordered domain labeled with Texas Red-DHPE is concentrated at the coverslip interface. By comparison, the fluorescence intensity of the binary GUVs did not decrease nearly as drastically in images taken above the coverslip, Figure 4B. Additionally, there was no evidence of membrane phase separation at the contacts between binary GUVs and the coverslip surface. Collectively, these observations suggest that the DOTAP-rich liquid disordered phase of the ternary GUVs orients itself toward the anionic surface of the glass coverslip, and that this contact site may promote local membrane phase separation. Considering that phase separation has been shown to promote drug release,^{28, 29} as discussed above, the results in Figure 4 provide qualitative support for the idea that combining cationic lipids with ternary lipid composition may help to promote drug delivery. Previous work from our lab has demonstrated that DOTAP-containing liposomes with ternary compositions can promote fusion with cellular membranes, resulting in the delivery of macromolecules to the cytoplasm.¹⁹ The present work utilizes similar liposomes to deliver doxorubicin, a small molecule drug. While the exact mechanism of doxorubicin release and cellular uptake remains unclear, our results in Figure 4 indicate the potential for membrane fusion between the contact sites of the liposomes and the cell surface.

Figure 4. Ternary liposomes containing DOTAP undergo phase separation upon contact with an anionic surface.

Evaluation of ternary GUVs (A) and binary GUVs (B) containing 3 mol% DOTAP at their point of contact with a glass coverslip. Confocal fluorescence microscopy images reveal cross-sections of GUVs at the coverslip and slightly above the coverslip. The white box in the confocal images at the coverslip represents the location of the inset, which displays an enlarged view of the contact area. The diagram on the right further explains the experimental setup. Line plots above the images display fluorescence intensity profiles across the GUVs as indicated by the white dotted lines. The fluorescence intensity in both plots was normalized to the maximum intensity of the confocal slice at the coverslip. Contrasts of the images were adjusted for visualization. All scale bars in example images indicate 5 μm; scale bars in insets indicate 2 μm.

Addition of DOTAP to ternary liposome compositions synergistically enhances efficacy of liposome-encapsulated doxorubicin.

We next evaluated the ability of DOTAP to enhance the delivery and cytotoxicity of liposomal doxorubicin. We concentrated doxorubicin inside of vesicles using a transmembrane ammonium sulfate gradient, as described above. Characterizing the liposomes via dynamic light scattering (DLS) before and after doxorubicin loading indicated that the diameter of the liposomes did not change substantially during drug loading, Supplement Figure S4. All DOTAP-containing membrane compositions retained a significant fraction of doxorubicin after dialysis, indicating that doxorubicin was concentrated within the liposomes, Figure 5A. Additionally, the substantial retention of doxorubicin within the liposomes after overnight dialysis suggests that passive release of drugs from the liposomes occurs over many hours, substantially longer than the two hour treatment window employed in the cell studies. To visualize doxorubicin delivery, we imaged HeLa cells by confocal fluorescence microscopy two hours after incubation with doxorubicin-loaded vesicles. Doxorubicin fluorescence was excited at 488 nm and its emission was detected through a bandpass emission filter centered at 525 nm (see methods). Doxorubicin fluorescence was clearly visible in the nuclei of cells treated with doxorubicin-loaded liposomes, Figure 5B. In particular, for a fixed concentration of DOTAP, the nuclei of cells treated with doxorubicin-loaded ternary liposomes displayed more intense doxorubicin fluorescence compared to the nuclei of cells treated with binary liposomes loaded with the same amount of doxorubicin. Increased doxorubicin fluorescence qualitatively implies that more doxorubicin entered the cell. Furthermore, the nuclei of cells treated with ternary liposomes that contained 5 mol% DOTAP displayed substantially brighter doxorubicin fluorescence in comparison to ternary liposomes that contained 0 mol% and 3 mol% DOTAP, Figure 5B. We determined that the doxorubicin fluorescence resided mainly in the cell nucleus based on corresponding brightfield images of the cells after two hours of incubation with doxorubicin-loaded vesicles (Figure 5B and Supplement Figure S8.)

Figure 5. Incorporation of DOTAP in ternary liposomes enhances cellular delivery and cytotoxicity of liposomal doxorubicin.

(A) ratio of final to initial doxorubicin content after overnight dialysis of all liposome formulations and for the unencapsulated drug (drastic reduction highlighted by red arrow). The black and white squares represent individual doxorubicin loading trials, and the gray bars represent the average of the two trials. (B) Example confocal fluorescence microscopy images comparing HeLa cells after two hours of incubation with doxorubicin-loaded binary and ternary liposomes containing 0 mol%, 3 mol%, and 5 mol% DOTAP. Scale bars represent 10 μm. The dashed white lines reveal the edges of the cells. (C – E) Dose-response curves depicting the percentage of nonviable cells upon delivery of increasing equivalent doxorubicin concentrations encapsulated within liposomes that contained 3 mol%, 5 mol%, and 10 mol% DOTAP. Open circles represent ternary liposomes and black triangles represent binary liposomes. The green lines (ternary) and blue lines (binary) indicate average percent nonviable cells from three independent trials of at least 5,000 cells per trial. (F) Bar chart comparing the percentage of nonviable cells upon delivery of 100 μM doxorubicin with binary and ternary liposomes containing 0 mol%, 3 mol%, 5 mol%, and 10 mol% DOTAP. The final bar indicates the percentage of nonviable cells upon delivery of 100 μM free doxorubicin. (G) Bar chart displaying the fold-change of nonviable cells upon delivery with each ternary liposome formulation relative to its respective binary formulation. Error bars indicate standard deviation between three independent trials.

To evaluate the cytotoxicity of doxorubicin-loaded liposomes containing DOTAP, we utilized the flow cytometry-based viability assay described in Figure 2. We observed increased cell lethality upon delivery of increasing concentrations of doxorubicin encapsulated by ternary liposomes containing 3 mol% DOTAP, Figure 5C. Comparing Figures 5C and 2D, we observed that adding DOTAP to ternary liposomes led to greater cell lethality compared to ternary liposomes lacking DOTAP. Specifically, at 100 μM equivalent doxorubicin concentration, 45% ± 1% of cells incubated with ternary liposomes containing 3 mol% DOTAP were nonviable, in comparison to 25% ± 2% of cells incubated with ternary liposomes that lacked DOTAP, Figure 5F. In contrast, adding 3 mol% DOTAP to binary liposomes did not substantially increase the cytotoxicity of encapsulated doxorubicin, rendering only 8% ± 1% of cells nonviable, Figure 5F. Summarizing these results, when 3 mol% DOTAP was included in the lipid composition, doxorubicin encapsulated within ternary liposomes was about 5-fold more cytotoxic than doxorubicin encapsulated within binary liposomes, Figure 5G. These results suggest that there is a synergistic relationship between inclusion of DOTAP and the ternary composition of the liposomes. While neither effect led to a substantial improvement in the cytotoxicity of doxorubicin on its own, the combination of the two effects in the same liposome increased the cytotoxicity of the doxorubicin to 65% of the unencapsulated drug, a 5-fold improvement over binary compositions with the same amount of DOTAP.

Next, we increased the concentration of DOTAP to 5 mol% to evaluate whether the synergistic effects we observed with liposomes containing 3 mol% DOTAP could be further increased. Using the same experimental procedures described above, we found that at each equivalent concentration of doxorubicin, liposomes containing 5 mol% DOTAP were more cytotoxic than liposomes that contained 3 mol% DOTAP, compare Figure 5C, D.

Comparing ternary and binary liposomes containing 5 mol% DOTAP, ternary liposomes were about 2-fold more cytotoxic than binary liposomes, reaching 81% of the performance of the free drug at an equivalent dose of 100 μM, Figure 5F, G. Together with results from liposomes containing 3 mol% DOTAP, these findings suggest that ternary composition and DOTAP content can be combined synergistically to enhance liposomal delivery. However, the relative increase in performance associated with the ternary composition was substantially larger, 5-fold versus 2-fold, for liposomes containing 3 mol% DOTAP compared to those containing 5 mol% DOTAP, Figure 5G. This result suggests that as DOTAP concentration increases, it drives increased delivery, whether it is part of a binary liposome or a ternary liposome. Therefore, the ability to use a ternary composition to reduce the requirement for DOTAP is confined to relatively low overall concentrations of the cationic lipid.

To further assess the effects of increasing DOTAP concentration, we evaluated cytotoxicity of HeLa cells incubated with liposomes containing 10 mol% DOTAP, Figure 5E. Again, using the same experimental procedures described above, we found that at each increasing equivalent concentration of doxorubicin, liposomes containing 10 mol% DOTAP were even more efficient at killing cells than liposomes containing 3 mol% and 5 mol% DOTAP, compare Figures 5C, D and E. At 100 μM of equivalent doxorubicin concentration, ternary liposomes containing 10 mol% DOTAP achieved 86% of the performance of the unencapsulated drug, Figure 5F. Despite the overall improvement in cytotoxicity upon increasing to 10 mol% DOTAP, relative performance associated with the ternary composition was reduced, Figure 5G. In fact, for liposomes containing 10 mol% DOTAP, binary liposomes somewhat outperformed ternary liposomes, Figure 5G. A possible explanation for the enhancement of doxorubicin delivery by binary vesicles containing DOTAP is likely that DOTAP itself is a fusogenic lipid known to promote delivery of macromolecules.^{19, 38} Our results demonstrate that increasing DOTAP concentration increases doxorubicin delivery to cells, regardless of whether the compositions are binary or ternary. However, for liposomes with relatively low DOTAP content, ternary compositions outperform binary compositions. Specifically, liposomes containing 3 mol% DOTAP show the largest improvement in delivery when moving from a binary to a ternary lipid composition. Notably, the enhanced performance of binary liposomes is unlikely to arise directly from membrane phase separation owing to insufficient unsaturated lipid content.¹⁵ Further, as shown in Figure 3A, these liposomes do not display phase separation. However, heterogeneity in lipid composition across the liposome population, coupled with lipid sorting at the cell liposome interface, could trigger phase separation in some liposomes, assisting delivery. Our results illustrate that there exists an optimal level of DOTAP content for which the ability of the ternary lipid composition to mediate doxorubicin delivery is maximized.

CONCLUSION

Here we demonstrate that ternary membrane compositions can work synergistically with cationic lipids to enhance the cytotoxicity of doxorubicin in liposomal formulations. Initially, we set out to determine if ternary composition, in the absence of cationic lipids, could enhance cytotoxicity of liposomal doxorubicin relative to liposomes with conventional binary compositions. Upon delivery to cells in vitro, we found that ternary composition alone did not substantially enhance cytotoxicity, Figure 2D. However, upon addition of DOTAP to ternary vesicles, a corresponding increase in cytotoxicity occurred. Specifically, we found that treatment of cells with ternary liposomes containing 3 mol% DOTAP enhanced cytotoxicity by over 5-fold relative to binary liposomes containing the same amount of DOTAP, Figure 5G. These results indicate a synergistic relationship between ternary compositions and DOTAP.

The precise mechanistic origin of this synergistic effect is not revealed by our studies. However, recent experiments demonstrating the ability of contact with a substrate to drive membrane phase separation provide a plausible mechanism.^{31, 32} In particular, we speculate that charge-mediated interactions between cationic ternary liposomes and the anionic cellular surface are capable of driving local membrane phase separation. This speculation is supported in part by our observation that DOTAP-containing liquid disordered phases appear to orient themselves toward the anionic surfaces of glass coverslips, resulting in local membrane phase separation at the site of contact with the coverslip, Figure 4A. Membrane phase boundaries are known to promote drug release,^{28, 29, 47} an effect which likely contributes synergistically to the well-documented ability of cationic lipids to promote lipid mixing and membrane fusion.^{19, 48–52} However, once positively charged liposomes have adhered to the negatively charged cellular surface, passive diffusion of drug molecules across the liposome membrane and cell membrane is also a possible mechanism of doxorubicin delivery to cells.⁹

The combination of ternary lipid mixtures with modest amounts of cationic lipids shows promise in overcoming the obstacle of efficient delivery of chemotherapeutics to the cellular interior. Our results indicate that enhanced delivery to cells can be achieved with as little as 3 mol% DOTAP, a significant reduction compared to 50 – 100 mol% reported in other DOTAP-based delivery systems.^{53, 54} Further, our results suggest that the overall concentration of DOTAP, or similar fusogenic surfactants, can be reduced by using ternary lipid systems that are capable of undergoing lipid phase separation. Using these approaches to optimize drug delivery has the potential to help reduce the systemic toxicity that presently limits the clinical performance of chemotherapeutics.

MATERIALS AND METHODS

Chemical Reagents.

Sucrose, glucose, HEPES (4–2(2-hydroxymethyl)-1-piparazineethane-sulfonic acid), sodium chloride, and hydrochloric acid (HCl) were purchased from Fisher Scientific. Ammonium sulfate, sodium phosphate, potassium phosphate, potassium chloride, sodium bicarbonate, lectin protein from Canavalia ensiformis, ATTO-488 NHS-ester, TCEP (tris(2-carboxyethyl) phosphine hydrochloride), and doxorubicin were purchased from Sigma-Aldrich. Fetal bovine serum (FBS), trypsin, penicillin, streptomycin, L-glutamine, PBS (phosphate buffered saline), and DMEM (Dulbecco’s modified Eagle medium) were purchased from GE Healthcare. Texas Red-DHPE (Texas Red-1,2-Dihexadecanoyl-sn-glycero-3-phosphoethanolamine) was purchased from Thermofisher. Trypan blue was purchased from Life Technologies. DPPC (1,2 – dipalmitoyl-sn-glycero-3-phosphocholine), DOPC (1,2 dioleoyl-sn-glycero-3-phosphocholine), cholesterol (from ovine wool), DOTAP (1,2 dioleoyl--3-trimethylammonium-propane), and PEG2000-DPPE (1,2 dipalmitoyl-sn-glycerol-3-phosphoethanolamine-N-[Methoxy(Polyethylene glycol)-2000]) were all purchased from Avanti Polar Lipids (Alabaster, AL). LIVE/DEAD fixable violet dead cell stain kit was purchased from Thermo Fisher Scientific. All reagents were used without further purification.

Giant Unilamellar Vesicles.

Giant Unilamellar Vesicle (GUV) electroformation was performed according to published protocols.⁴⁷ The lipids, dissolved in chloroform, were spread on indium-tin-oxide (ITO) coated glass slides (resistance ~8–12 Ω-sq⁻¹) and were placed in a vacuum desiccator for at least 2 hours to remove all of the solvent. The vesicles were prepared using a 370-milliosmole sucrose solution. The electroformation oven was set to approximately 55°C to exceed the melting temperature of DPPC, 41°C. The voltage was increased every three minutes from 50 mVpp to 1400 mVpp for the first 30 minutes at a frequency of 10 Hz. The voltage was then held at 1400 mVpp for 120 minutes, and finally it was increased to 2200 mVpp for the last 30 minutes during which the frequency was adjusted to 5 Hz. After electroformation, the vesicle solution osmolarity was measured using a vapor pressure osmometer (Wescor).

Small Unilamellar Vesicles.

Small Unilamellar Vesicles (SUVs) were extruded using a mini-extruder (Avanti). First, lipids were mixed and dried using N₂ gas and placed under vacuum for a minimum of 2 hours. Lipid films were hydrated in 120mM, pH 5.3, 270 mOsm ammonium sulfate buffer for 30 minutes at 60°C, such that the lipid concentration was 20 mg/mL. Next, the hydrated lipid mixture was sonicated in a bath sonicator for 1 minute, and then extruded through 100-nm polycarbonate filters (VWR) for a minimum of 21 passes at 65°C. A Zetasizer Nano ZS (Malvern) was used for dynamic light scattering measurements.

Doxorubicin Loading.

To exchange ammonium sulfate, SUVs underwent two, 30-minute rounds of dialysis using 1000 kDa MWCO dialysis membranes (Spectrum Labs) in phosphate buffered saline (PBS) pH 7.4 at 4°C. Then, 650 μL of PBS, 250 μL of the 20 mg/mL extruded liposome solution, and 100 μL of 10 mg/mL of doxorubicin dissolved in dimethyl sulfoxide (DMSO) were mixed in a test tube for 1 hour at 40°C. The test tube was covered with parafilm to prevent evaporation. Next, this solution was dialyzed overnight using 1000 kDa MWCO dialysis membranes (Spectrum Labs) to remove unencapsulated doxorubicin. Absorbance measurements were taken using a Cytation 3 Multi-Mode Reader (BioTek) at the following time points: initial loading, 1 hour after loading at 40°C, and after overnight dialysis. Absorbance was measured at 480 nm and 485 nm, and the concentrations calculated from each wavelength were averaged to obtain a more accurate overall doxorubicin concentration. A Zetasizer Nano ZS (Malvern) was used for dynamic light scattering measurements.

Microscopy.

Spinning disc confocal microscopy (Zeiss Axio Observer Z1 with Yokagawa CSU-X1M) was used to image GUVs and cells. Laser wavelengths of 488-nm and 561-nm were used for excitation. The bandpass emission filters were centered at 525-nm with a 50-nm bandwidth, and 629-nm with a 62-nm bandwidth. Plan-Apochromat 100× 1.4 numerical aperture oil objective was used. A cooled (−70°C) EMCCD iXon3 897 camera was used for imaging (Andor Technology). A Zeiss AxioObserver microscope with 10x and 20x objectives was used for widefield imaging of cells. Cells were cultured on acid cleaned 22-mm square coverslips (Fisherbrand) for cell experiments using spinning disc confocal microscopy.

Acid Cleaning Coverslips.

22 × 22 mm glass coverslips (Fisherbrand) were heated at 60°C in 1M HCl in a covered glass beaker for approximately 10 hours and allowed to cool to room temperature. Distilled water was used to rinse coverslips. Next, coverslips were placed in distilled water in a covered glass beaker and sonicated in a bath sonicator for 15 minutes 3 times. After sonication in distilled water, the coverslips were sonicated in 50% ethanol and 50% distilled water for 15 minutes, then in 70% ethanol and 30% distilled water for 15 minutes, and finally 95% ethanol and 5% distilled water for 15 minutes. Coverslips were then stored in 95% ethanol and 5% distilled water.

Preparation of Slides for Imaging of Giant Vesicles.

Giant Unilamellar Vesicles were diluted by a factor of 4 in phosphate buffered saline (PBS) pH 7.4 raised to a final osmolarity of approximately 340-milliosmole by addition of sucrose. The slight osmotic tension suppressed membrane fluctuations, improving image quality. In all cases vesicles were observed in small, sealed, disposable chambers composed of 24 × 40 mm glass cover slips (Fisherbrand) and spacers made from 3 layers of double-sided tape. The samples were prepared approximately 5 minutes prior to imaging to allow GUVs to settle to the bottom of the coverslip.

Lectin Protein Labeling.

Lectin from Canavalia ensiformis was labeled with ATTO-488 NHS-ester in a buffer consisting of 150 mM sodium bicarbonate, 20 mM KCl, and 1 mM TCEP (pH 8.23). ATTO-488 dye was added to the lectin protein at a stoichiometric ratio of 8 dye molecules per protein molecule and allowed to react for at least 30 minutes at room temperature. The resulting labeling ratio for the lectin protein was about 1 dye molecule per protein. Unreacted dye was removed using Centri-Spin size exclusion columns (Princeton Separations). Protein and dye concentrations were measured using UV-vis spectroscopy.

Imaging GUVs with Lectin Protein.

Giant Unilamellar Vesicles composed of binary and ternary compositions containing 24 mol% DOTAP were diluted by a factor of 5 in a 20 mM HEPES and 50 mM sodium chloride buffer (pH 7.4). Prior to addition, the buffer was raised to a final osmolarity of approximately 10% less than that of the GUV samples so that the osmotic tension enables visualization of spherical vesicles. 2 μM of ATTO-488 labeled lectin protein was then added to the diluted GUV sample, and the resulting mixture was imaged in small, sealed, disposable chambers composed of 24 × 40 mm glass cover slips (Fisherbrand) as described above.

Imaging GUVs at Coverslip Interface.

Giant Unilamellar Vesicles were diluted by a factor of 5 in phosphate buffered saline (PBS) pH 7.4 raised to a final osmolarity of approximately 370-milliosmole by addition of glucose. The osmolarity should be the same or slightly higher than that of the GUV samples so little osmotic tension is present to prevent immediate rupturing of GUVs on the glass. In all cases, vesicles were observed in small, sealed, disposable chambers made of 24 × 40 mm glass cover slips (Fisherbrand) as described above.

Estimating the Area Fraction of Unsaturated-rich Lipid Domains.

The average area of the liquid-disordered domain as a percentage of total vesicle surface area was estimated using 3D confocal reconstructions of the vesicle. From these reconstructions the domain was considered to be a spherical cap for which the area was calculated based on measurements of the vesicle diameter and the cap diameter.

Cell Culture.

HeLa cells were purchased from American Type Culture Collection (ATCC). All cells were cultured in DMEM high glucose supplemented with 10% FBS and 1% penicillin, streptomycin, and L-glutamine (PSLG). All cells were incubated at 37°C with 5% CO₂ and passaged every 48–72 hours. For fluorescence microscopy, cells were grown on acid-cleaned 22 mm × 22 mm glass coverslips (Fisherbrand) in 6-well plates (Corning) for 24 hours. For flow cytometry, cells were grown in 48-well plates (Corning) for 24 hours.

Doxorubicin Cytotoxicity Determined by Flow Cytometry.

HeLa cells were plated in a 48 well plate at a density of 30,000 cells per well and a total media volume of 500 μL media per well, one day prior to doxorubicin delivery. For each delivery trial, a batch of doxorubicin loaded liposomes was prepared as described above. The quantity of doxorubicin encapsulated within each batch of liposomes was determined as described above. Liposomes were diluted in media to achieve 10 μM, 50 μM, and 100 μM doses of doxorubicin, and then each dose was added to the HeLa cells. Cells were then incubated in the dark at 37 °C with 5% CO₂ for 2 hours. After 2 hours of incubation, the HeLa cells were rinsed with 500 μL media and then incubated in 500 μL fresh media at 37 °C with 5% CO₂. After a total of 24 hours, HeLa cell viability was analyzed using a LIVE/DEAD fixable violet dead cell stain (Thermo Fisher). This stain works by reacting with free amine groups on or within cells. When the cell membrane is compromised, the stain further reacts with amine groups on the cellular interior, resulting in much more staining and an increase in fluorescence. Thus, live and dead cell populations can be determined based on the intensity of staining. For the violet dead cell stain assay, the media from each well of cells was transferred into a fresh 1.5 mL Eppendorf tube. The remaining cells were then detached from the 48-well plate with 150 μL of trypsin for 5 minutes at 37 °C, 5% CO2. Trypsinized cells were then added to the Eppendorf tube that contained the media aspirated from their specific well. Cells within this combined sample were subsequently pelleted for 5 minutes at 300 × g. The resulting cell pellet was resuspended in 500 μL of PBS, and 0.5 μL of violet dead cell stain was added to each tube and mixed well. Upon addition of the stain, the samples were covered and incubated on ice for 30 min. After 30 minutes, the samples were spun again at 300 × g for 5 minutes, and the resulting pellet of cells were resuspended in 300 μl of PBS in preparation for flow cytometry analysis. A Guava easyCyte Flow Cytometer (Millipore Sigma) with 405, 488, and 532 nm excitation lasers was used for all flow cytometry experiments. All data were collected at 35 μL/min. Flow cytometry data were analyzed using FlowJo (Treestar). A rectangular gate was drawn in forward scattering versus side scattering plots to exclude debris and contain the majority of cells (Supplement Figure S3). Within this gate, cell populations were further analyzed by plotting histograms of violet stain intensity to determine live versus dead cells. To determine the percentage of non-viable cells in each sample, a threshold was drawn on the flow cytometry fluorescence histograms as close to the minimum point between the two peaks in the histogram of violet dead cell stain intensity as possible (Supplement Figure S5 and S7). The percentage of cells with fluorescence below the threshold were considered viable, and the percentage of cells with fluorescence above these thresholds were considered non-viable.